[TB3] SLAM with the TurtleBot3

TurtleBot
1

Hello, :slight_smile:

Have you seen IEEE Spectrum’s video Friday last? A video with Turtlebot3 was introduced. We will share the development news of Turtlebot3 every week. :wink:

http://spectrum.ieee.org/automaton/robotics/robotics-hardware/video-friday-kengoro-sweaty-robot-fotokite-camera-drone-yann-lecun-ai

Today, I would like to introduce the video related to SLAM which is the basic function of TurtleBot3.
Even if it is small size, low cost, we will do our best to SLAM and Navigation which is the basic function of Turtlebot brand.

Date: 2016.11.29
Robot: TurtleBot3 basic model
Sensor: Laser Distance Sensor
ROS package for SLAM: Gmapping / Cartographer
Place: ROBOTIS Labs & HQ, 15th-floor corridor
Duration: 55 minutes
Total travel distance: 351 meters

If you have any questions or concerns for TurtleBot3, please notice us.
We hope to discuss it with you.

Best regards,
Pyo

2 Likes
2

Could you share the associated bag for this video? I think the Cartographer results could be easily improved and we’d be happy to host a TB3 configuration with a demo bag.

3

You mentioned that you are using a “Laser Distance Sensor”, can you share which one you are using specifically?

It would be interesting to know how well mapping performs with a lower FOV sensor, like an Astra or Realsense.

Also is this running in real-time, if so which SBC are you using?

4

@damonkohler

Hi Damon Kohler,

We are still lacking, but I think we can do better if we get your help.
We share our bag data. https://goo.gl/7QAudM

Thanks :slight_smile:

5

@rohbotics

We are using a variety of LDS. Typically, these sensors are RPLidar A1 and A2, SICK, Hokuyo.
Also, we have recently found a LDS that is cheaper, but satisfactory in performance. It is in testing.
It will be released soon.

We will support various SBCs for TurtleBot3.
Among them, we will use Raspberry Pi for the basic version of TurtleBot3 and Intel Joule for the premium version of TurtleBot3.

6

The bag doesn’t seem to contain odometry. I assume you had that available for gmapping, though? Cartographer would also benefit from it.

7

@damonkohler
Thank you for your comments, I think that you can get the odometry information from /tf of above bag data.
Actually, we are currently using Gmapping as our main SLAM package. But we will soon change to the Cartographer package. You will need your help. :slight_smile:

8

Thanks, Pyo. Glad to hear it!

Odometry is usually provided via nav_msgs/Odometry messages. The /odom frame in /tf is then a product of some node that processes the odometry messages.

9

Thank you.

We are awaiting the Kinetic version of the cartographer. It is very difficult to install the current indigo version on ROS Kinetic. It is not easy due to the problem of ceres-solver and errors during installation. So we were using the modified package of the link below for ROS kinetic. We are hoping to be able to install the Google Cartographer as binary from ROS kinetic soon.

Https://github.com/hitcm/ceres-solver-1.11.0.Git


10

We’ll work on some documentation for installing from source on Kinetic. But, having a binary release is blocked on having an updated Ceres package for Kinetic. I believe @tfoote is working on that.

Can you add a nav_msgs/Odometry publisher?

11

I saw your thread. Ceres Package?
I also believe @tfoote is working on that. :wink:

And, I will add an odometry publisher you mentioned into the next update as soon as possible.

Thanks,